Mechanism-guided wear severity assessment of worn surfaces with multiple damages

The wear severity of mechanical components is commonly evaluated through visual inspection of worn surfaces during routine maintenance. However, the accuracy of the intelligent assessment is hampered by the differences and superposition of co-existing damages. To address these issues, a wear severity assessment method is developed by integrating improved damage identification and state reasoning. Initially, the co-existing damages are identified with a stepwise refinement identification model incorporating spatial and categorical guidance. These identified damages are then quantified by comparing the worn surface to its initial state. On this basis, a mixture-of-experts model, utilizing an indicator-damage-state reasoning framework, is developed to deduce the overall wear severity and contributions of individual damages. With the measures, the intelligent wear severity assessment is extended to complex worn surfaces involving multiple damages. This method is validated using worn surface samples sourced from the main-shaft bearings of aero engines, which involve abrasive, fatigue, adhesive, and corrosion wear. The results reveal that the diagnosis accuracy of wear severity has been improved from 87.7 % to 93.8 % compared to the existing methods. Furthermore, the proposed method is employed to rolling-sliding friction tests, where it performs excellently in diagnosing wear mechanisms and severities of the test roller and ring.